mistralrs-core 0.8.1

Fast, flexible LLM inference.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69

mod bs1770;
mod dia;
pub mod utils;

use std::{str::FromStr, sync::Arc};

pub use dia::{DiaConfig, DiaPipeline};
use serde::{Deserialize, Serialize};

#[derive(Clone, Copy, Debug, Deserialize, Serialize, PartialEq)]
pub enum SpeechLoaderType {
    #[serde(rename = "dia")]
    Dia,
}

impl FromStr for SpeechLoaderType {
    type Err = String;
    fn from_str(s: &str) -> Result<Self, Self::Err> {
        match s {
            "dia" => Ok(Self::Dia),
            a => Err(format!(
                "Unknown architecture `{a}`. Possible architectures: `dia`."
            )),
        }
    }
}

impl SpeechLoaderType {
    /// Auto-detect speech loader type from a config.json string.
    /// Extend this when adding new speech pipelines.
    pub fn auto_detect_from_config(config: &str) -> Option<Self> {
        if serde_json::from_str::<DiaConfig>(config).is_ok() {
            return Some(Self::Dia);
        }
        None
    }
}

#[derive(Clone, Copy, Debug)]
pub enum SpeechGenerationConfig {
    Dia {
        max_tokens: Option<usize>,
        cfg_scale: f32,
        temperature: f32,
        top_p: f32,
        top_k: Option<usize>,
    },
}

impl SpeechGenerationConfig {
    pub fn default(ty: SpeechLoaderType) -> Self {
        match ty {
            SpeechLoaderType::Dia => Self::Dia {
                max_tokens: None,
                cfg_scale: 3.,
                temperature: 1.3,
                top_p: 0.95,
                top_k: Some(35),
            },
        }
    }
}

#[derive(Clone, Debug)]
pub struct SpeechGenerationOutput {
    pub pcm: Arc<Vec<f32>>,
    pub rate: usize,
    pub channels: usize,
}